Results for "vector representation"

Set of vectors closed under addition and scalar multiplication.

A datastore optimized for similarity search over embeddings, enabling semantic retrieval at scale.

Measure of vector magnitude; used in regularization and optimization.

Mathematical foundation for ML involving vector spaces, matrices, and linear transformations.

Internal representation of the agent itself.

Vector whose direction remains unchanged under linear transformation.

A continuous vector encoding of an item (word, image, user) such that semantic similarity corresponds to geometric closeness.

Converts logits to probabilities by exponentiation and normalization; common in classification and LMs.

Automatically learning useful internal features (latent variables) that capture salient structure for downstream tasks.

Architecture that retrieves relevant documents (e.g., from a vector DB) and conditions generation on them to reduce hallucinations.

Raw model outputs before converting to probabilities; manipulated during decoding and calibration.

Mechanisms for retaining context across turns/sessions: scratchpads, vector memories, structured stores.

Optimal estimator for linear dynamic systems.

Vectors with zero inner product; implies independence.

Matrix of first-order derivatives for vector-valued functions.

Visualization of optimization landscape.

Structured graph encoding facts as entity–relation–entity triples.

Diffusion performed in latent space for efficiency.

Model that compresses input into latent space and reconstructs it.

Mathematical representation of friction forces.

Internal representation of environment layout.

Learning a function from input-output pairs (labeled data), optimizing performance on predicting outputs for unseen inputs.

A structured collection of examples used to train/evaluate models; quality, bias, and coverage often dominate outcomes.

A parameterized mapping from inputs to outputs; includes architecture + learned parameters.

A parameterized function composed of interconnected units organized in layers with nonlinear activations.

Gradients grow too large, causing divergence; mitigated by clipping, normalization, careful init.

Networks with recurrent connections for sequences; largely supplanted by Transformers for many tasks.

Retrieval based on embedding similarity rather than keyword overlap, capturing paraphrases and related concepts.

Limiting gradient magnitude to prevent exploding gradients.

Matrix of second derivatives describing local curvature of loss.